RDS snapshot export in Amazon S3

When an RDS snapshot export to S3 gets stuck in the "Running" state for an extended period, this is a known issue that can occur even with relatively small databases like your 11 GB MySQL Aurora instance.

The export process typically involves several phases - first restoring and scaling the database, then extracting data to Amazon S3. During this process, you might see the task progress reach a high percentage (often 99%) and then appear to stall. The fact that you're seeing files being created and then deleted in your S3 bucket suggests the process is still working, but may be encountering issues.

Several factors can affect the export time:

• Data characteristics: Tables with well-distributed numeric primary key or index columns export faster than those with string-based index columns or without suitable partitioning columns
• The export process first needs to restore the entire database before extracting data

Here are some troubleshooting steps you can take:

1. Check that your IAM role has the proper permissions and trust relationship. The role must allow the "export.rds.amazonaws.com" service principal to assume it.

2. Verify that your S3 bucket is in the same region as your RDS instance, as cross-region exports can cause issues.

3. Check for any S3 bucket policy restrictions that might be interfering with the export process.

4. Monitor the export task using the AWS Management Console, AWS CLI (describe-export-tasks command), or RDS API to get more detailed information about the status.

If the issue persists, you might consider:

• Canceling the current export task and initiating a new one
• Using AWS Database Migration Service (DMS) as an alternative method to export your RDS data to S3
• Contacting AWS Support for more detailed diagnostics specific to your account and setup

The "Delete marker" you're seeing for the test files is likely part of the normal process where the service verifies write permissions to the S3 bucket before proceeding with the actual export.
Sources
RDS Snapshot Export to S3 Stuck at 99% In Progress | AWS re:Post
AWS RDS export to s3, is giving me a very wrong error, that is, my s3 bucket is not in same region | AWS re:Post
i want to export rds db to the amazon s3 but facing this issue i already created the role export-rds-db-s3 with s3 full access and glue console full access | AWS re:Post

The extended "Starting" state for your RDS snapshot export is indeed normal during AWS's validation phase, but after 12+ hours, it's worth verifying a few specific configurations that commonly cause delays. Ensure your IAM role's trust policy explicitly allows export.rds.amazonaws.com to assume it, and confirm the S3 bucket policy grants the role s3:PutObject without restrictive conditions (especially if using bucket encryption with KMS, which requires additional kms:GenerateDataKey and kms:Decrypt permissions). Also, check that the S3 bucket is in the same region as the snapshot and doesn't have object lock enabled. The temporary test files with delete markers confirm basic permissions are working, but the prolonged initialization often stems from internal AWS capacity allocation—wait up to 24 hours total before escalating. If still stuck, use the AWS CLI to cancel (aws rds cancel-export-task) and retry during off-peak hours, or open a support case (even on basic support, you can use the AWS Management Console's "Support Center" for technical issues). For future exports, consider splitting large snapshots into smaller ones if feasible, as exports over 1 TB frequently experience longer initialization times.